On 9/25/23 19:20, Jinoh Kang wrote: > As an outsider's perspective, I think this kind of thing is where selftests > really shine. I got the impression that Xen will need to rely on numerous > other > platform oddities, the documentation of which are often unavailable. > > Of course, adding a whole new test infrastructure in code freeze is not > viable. > Maybe I have missed something, but I only see three paths forward here: > > 1. Reference the most relevant paragraph in SDM/APM, but don't quote it. > Keep the current explanation, and state that the manual is vague anyway. > > 2. Acknowledge that SDM/APM is incomplete, and completely abandon the manual > as the *authoritative* source of information. Perhaps embed a sample test > program that demonstrates the behavior, if it isn't too long. > > 3. Actually assert in runtime that DR6 behaves as expected. >
Just a heads-up; has there been any progress on this part? Please let me know if you need anything. I'm happy to help! -- Sincerely, Jinoh Kang